Re: Review Request 18341: Modified the Registrar to notify callers only _after_ the storage operation completes.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/#review35536 --- Ship it! src/master/registrar.cpp https://reviews.apache.org/r/18341/#comment66132 It looks like 'applied' is true iff 'operator ()' returns Some or None. Can we simplify the semantics and not require setting 'applied'? What about another wrapper that calls 'operator ()' and sets 'applied' based on the result of 'operator ()'. Also, do you perceive any cases of where Error won't turn into an 'applied = false' case? Finally, since maybe s/applied/success/ since some operations aren't technically applied as discussed in Vinod's previous review. src/master/registrar.cpp https://reviews.apache.org/r/18341/#comment66130 Looks like mis-alignment here. src/master/registrar.cpp https://reviews.apache.org/r/18341/#comment66129 Minor nit: why ERROR instead of WARNING? This isn't an error, this is an expected case! Also, when first reading this I was sort of surprised that we were just printing this out and not completing the promises/operations here. Perhaps a brief comment explaining that we set this first and do the same for the operations below? - Benjamin Hindman On Feb. 26, 2014, 2:19 a.m., Ben Mahler wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/ --- (Updated Feb. 26, 2014, 2:19 a.m.) Review request for mesos, Benjamin Hindman and Vinod Kone. Bugs: MESOS-764 https://issues.apache.org/jira/browse/MESOS-764 Repository: mesos-git Description --- We previously were setting the Promises in some cases _before_ writing to storage. While this approach was correct, it requires some time and explanation to convince oneself of this. This is why this patch simplifies the semantics by only setting the Promises _after_ we've written to storage. Also, 'Mutation' has now become 'Operation' to better reflect the fact that some operations do not mutate state when applied. Diffs - src/master/registrar.cpp ee16121035db21d966ee151483dd23cbc4a495c2 Diff: https://reviews.apache.org/r/18341/diff/ Testing --- make check Thanks, Ben Mahler
Re: Path forward to C++11.
Update on MESOS-750: 0.17.0 is being released with the --with-cxx11 capability at configure time. I chatted with some of the committers and we'd like to move forward with the final two phases of MESOS-750 as follows: 0.18.0: --with-cxx11 will become the _default_. --without-cxx11 will be available for those who wish to use older compilers. 0.19.0: c++11 required; Mesos will not build with a compiler that does not have c++11 support. If there are any objections please speak now. If not, I'll update the ticket to reflect the current path towards C++11. On Mon, Oct 28, 2013 at 8:09 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Thanks for the heads up, we're likely to upgrade to 3.3.6, but I've yet to confirm that it builds with c++11. On Thu, Oct 24, 2013 at 8:10 AM, Tim St Clair tstcl...@redhat.com wrote: +1 to the approach. FWIW made a quick update to the zookeeper.spec ( http://koji.fedoraproject.org/koji/buildinfo?buildID=465529) and updated: export CC='gcc -std=c11' export CXX='g++ -std=c++11' prior to 'configure', and it failed compilation pretty quickly. It may make sense to track integration issues and versions...somewhere. So when --with-cpp11 is enabled, it can hard check on version = fixed_version. Cheers, Tim - Original Message - From: Benjamin Mahler benjamin.mah...@gmail.com To: dev@mesos.apache.org Sent: Wednesday, October 23, 2013 4:46:49 PM Subject: Path forward to C++11. As discussed in MESOS-750 https://issues.apache.org/jira/browse/MESOS-750, we would like to move to C++11. The proposed plan is to do this in phases to allow everyone time to adjust to this change: Phase 1: Send an email to the dev@ list soliciting objections / feedback. We're currently here. :) Phase 2: Add an option --with-cpp11 that defaults to false. Phase 3: At this point, developers can and should begin using --with-cpp11. In this phase, we'll need to make sure the project and dependencies can compile with c++11. Phase 4: Default to --with-cpp11 but still leave the option in order to allow folks to turn this off if needed. At this point, we still cannot rely on C++11. Phase 5: Remove --with-cpp11 leaving no ability to turn it off. From this point forward, we can begin to use C++11 features unconditionally. Please reply to this thread with any objections, comments, or advice! To get a better sense of how others feel about this (I currently only know that some of the other core contributors are on board), please feel free to reply with something as simple as a +1 as well. Thanks! Ben -- Cheers, Tim
Re: Review Request 18403: Added support for launching tasks by TaskInfo.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18403/#review35504 --- I think we need to work on this a little. It seems that making the launching state explicit and then cleaning up behavior (killTask) when in this state is necessary. src/slave/containerizer/containerizer.hpp https://reviews.apache.org/r/18403/#comment66140 Should this be in mesos_containerizer? I was thinking if the task didn't have task.executor() then the determination of the executor would be containerizer implementation dependent. src/slave/containerizer/mesos_containerizer.cpp https://reviews.apache.org/r/18403/#comment66141 Why is this a Future? Can you not just pass the ExecutorInfo to exec() and have it return? src/slave/containerizer/mesos_containerizer.cpp https://reviews.apache.org/r/18403/#comment66143 const ExecutorInfo ? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66056 Could you swap this logic? i.e., if (contains) { launch = framework-launchingExecutors[]; } else { launch = framework-launch(); } src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66057 What happens to Executor* if the future is ready? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66058 s/rebust/robust/ src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66060 Previously executor == NULL was unexpected but now it seems weird to have NULL as a 'normal' value. Can we capture this is a better way? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66062 We shouldn't be returning TASK_LOST and claiming we cannot find the executor here if the executor is actually launching. If we haven't sent a launching executor the task then we should prevent that and act appropriately. src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66063 What about introducing a new executor state - Executor::LAUNCHING? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66065 s/We might be in a situation where the executor is launching but not yet ready. The second half is scheduled until executor is ready, or called directly if executor is running./If the executor is still launching we'll defer registering until it is ready./? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66078 ditto - what happens to Executor* if the future is ready? src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66134 ditto - Executor* src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66135 okay :-p src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66136 perhaps not in this change, but this should done when the FutureExecutorInfo is ready rather than now. src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66137 ditto - should start timeout when the launch is ready. src/slave/slave.cpp https://reviews.apache.org/r/18403/#comment66138 The determination of executor resources should be done by each containerizer implementation if they choose an executor. - Ian Downes On Feb. 26, 2014, 12:04 a.m., Niklas Nielsen wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18403/ --- (Updated Feb. 26, 2014, 12:04 a.m.) Review request for mesos, Adam B, Benjamin Hindman, Ben Mahler, Ian Downes, Till Toenshoff, and Vinod Kone. Bugs: MESOS-922 https://issues.apache.org/jira/browse/MESOS-922 Repository: mesos-git Description --- This patch delegates the choice of executor to the containerizer by removing executorInfo dependencies up until Containerizer::launch(). Containerizer::launch() now returns a future to the executor info that is being run and the slave creates the corresponding executor structure when launch completes. This means message handling from the running executor to the slave in the interim where the executor structure has not created, need to be enqueued until executor is ready. So far, registerExecutor() and reregisterExecutor() has been split into two continuations to deal with this issue. Diffs - src/slave/containerizer/containerizer.hpp d9ae326 src/slave/containerizer/containerizer.cpp d0a1023 src/slave/containerizer/mesos_containerizer.hpp ee1fd30 src/slave/containerizer/mesos_containerizer.cpp 6d990cb src/slave/slave.hpp 01b80df src/slave/slave.cpp 4f5349b src/tests/containerizer.hpp 5686398 src/tests/containerizer.cpp bfb9341 Diff: https://reviews.apache.org/r/18403/diff/ Testing --- make check Thanks, Niklas Nielsen
Re: Path forward to C++11.
What do you mean by 'C++11 support' in the compiler? g++4.7 supports some features, not all. I think we need to define the set of C++11 features we want to support first before making any decisions regarding timeline. Then we can look at the state of the nation in regards to C++11 feature support in different compilers on different OSs, and which are available out of the box. We don't want to limit our community involvement by shutting out certain platforms for development. Ideally, we'd be able to say something like 0.19.0 builds on g++4.7 and clang 3.3 or greater and 0.20.0 builds on g++4.8 and clang 3.4 or greater instead of blanket C++11 support. On Wed, Feb 26, 2014 at 10:20 AM, Benjamin Mahler benjamin.mah...@gmail.com wrote: Update on MESOS-750: 0.17.0 is being released with the --with-cxx11 capability at configure time. I chatted with some of the committers and we'd like to move forward with the final two phases of MESOS-750 as follows: 0.18.0: --with-cxx11 will become the _default_. --without-cxx11 will be available for those who wish to use older compilers. 0.19.0: c++11 required; Mesos will not build with a compiler that does not have c++11 support. If there are any objections please speak now. If not, I'll update the ticket to reflect the current path towards C++11. On Mon, Oct 28, 2013 at 8:09 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Thanks for the heads up, we're likely to upgrade to 3.3.6, but I've yet to confirm that it builds with c++11. On Thu, Oct 24, 2013 at 8:10 AM, Tim St Clair tstcl...@redhat.com wrote: +1 to the approach. FWIW made a quick update to the zookeeper.spec ( http://koji.fedoraproject.org/koji/buildinfo?buildID=465529) and updated: export CC='gcc -std=c11' export CXX='g++ -std=c++11' prior to 'configure', and it failed compilation pretty quickly. It may make sense to track integration issues and versions...somewhere. So when --with-cpp11 is enabled, it can hard check on version = fixed_version. Cheers, Tim - Original Message - From: Benjamin Mahler benjamin.mah...@gmail.com To: dev@mesos.apache.org Sent: Wednesday, October 23, 2013 4:46:49 PM Subject: Path forward to C++11. As discussed in MESOS-750 https://issues.apache.org/jira/browse/MESOS-750, we would like to move to C++11. The proposed plan is to do this in phases to allow everyone time to adjust to this change: Phase 1: Send an email to the dev@ list soliciting objections / feedback. We're currently here. :) Phase 2: Add an option --with-cpp11 that defaults to false. Phase 3: At this point, developers can and should begin using --with-cpp11. In this phase, we'll need to make sure the project and dependencies can compile with c++11. Phase 4: Default to --with-cpp11 but still leave the option in order to allow folks to turn this off if needed. At this point, we still cannot rely on C++11. Phase 5: Remove --with-cpp11 leaving no ability to turn it off. From this point forward, we can begin to use C++11 features unconditionally. Please reply to this thread with any objections, comments, or advice! To get a better sense of how others feel about this (I currently only know that some of the other core contributors are on board), please feel free to reply with something as simple as a +1 as well. Thanks! Ben -- Cheers, Tim -- Dominic Hamon | @mrdo | Twitter *There are no bad ideas; only good ideas that go horribly wrong.*
Re: Review Request 18361: Implemented MESOS-945: Added hostname attribute to framework, display it in master's web ui
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18361/#review35560 --- Ship it! Ship It! - Vinod Kone On Feb. 25, 2014, 2:16 p.m., Bernd Mathiske wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18361/ --- (Updated Feb. 25, 2014, 2:16 p.m.) Review request for mesos and Vinod Kone. Bugs: MESOS-945 https://issues.apache.org/jira/browse/MESOS-945 Repository: mesos-git Description --- - Added a field to FrameworkInfo in protobuf spec - Added table fields in html of web ui that display framework's hostname for running and terminated frameworks - Added line to scheduler constructor that sets framework hostname. This is just about the first centralized piece of code that every framework executes. - Refactored scheduler constructor so that code duplication is avoided. Diffs - include/mesos/mesos.proto 69a4a60 include/mesos/scheduler.hpp 2e4707e src/master/http.cpp c941274 src/sched/sched.cpp dcb3158 src/slave/http.cpp 7c4cfba src/webui/master/static/frameworks.html 8686b00 Diff: https://reviews.apache.org/r/18361/diff/ Testing --- Built (with Clang) and ran local Mesos on Mac. Ran C++ and Java test frameworks. Observed new hostname display in web UI. Built Mesos on Ubuntu 13.10 with gcc 4.8.1. Thanks, Bernd Mathiske
Re: Review Request 18341: Modified the Registrar to notify callers only _after_ the storage operation completes.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/ --- (Updated Feb. 26, 2014, 7:54 p.m.) Review request for mesos, Benjamin Hindman and Vinod Kone. Changes --- Updates per benh review. Bugs: MESOS-764 https://issues.apache.org/jira/browse/MESOS-764 Repository: mesos-git Description --- We previously were setting the Promises in some cases _before_ writing to storage. While this approach was correct, it requires some time and explanation to convince oneself of this. This is why this patch simplifies the semantics by only setting the Promises _after_ we've written to storage. Also, 'Mutation' has now become 'Operation' to better reflect the fact that some operations do not mutate state when applied. Diffs (updated) - src/master/registrar.cpp ee16121035db21d966ee151483dd23cbc4a495c2 Diff: https://reviews.apache.org/r/18341/diff/ Testing --- make check Thanks, Ben Mahler
Re: Review Request 18341: Modified the Registrar to notify callers only _after_ the storage operation completes.
On Feb. 26, 2014, 6:04 p.m., Benjamin Hindman wrote: src/master/registrar.cpp, lines 79-83 https://reviews.apache.org/r/18341/diff/3/?file=504150#file504150line79 It looks like 'applied' is true iff 'operator ()' returns Some or None. Can we simplify the semantics and not require setting 'applied'? What about another wrapper that calls 'operator ()' and sets 'applied' based on the result of 'operator ()'. Also, do you perceive any cases of where Error won't turn into an 'applied = false' case? Finally, since maybe s/applied/success/ since some operations aren't technically applied as discussed in Vinod's previous review. I've made some updates here. Now Operation is still a functor, but the () operator wraps a call to a protected 'apply' function. This is done in order to set 'success' based on the result of the 'apply' call. To answer your question, 'success' is true iff !apply().isError() and I think this will generalize to all operations. Now each Operation overrides the pure virtual apply() call and does not have to deal with setting the 'success' boolean, which is in private scope anyway and is thus hidden from the subclasses of Operation. Let me know how it looks, it's a bit unfortunate that we've introduced a copy across () and apply(). - Ben --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/#review35536 --- On Feb. 26, 2014, 2:19 a.m., Ben Mahler wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/ --- (Updated Feb. 26, 2014, 2:19 a.m.) Review request for mesos, Benjamin Hindman and Vinod Kone. Bugs: MESOS-764 https://issues.apache.org/jira/browse/MESOS-764 Repository: mesos-git Description --- We previously were setting the Promises in some cases _before_ writing to storage. While this approach was correct, it requires some time and explanation to convince oneself of this. This is why this patch simplifies the semantics by only setting the Promises _after_ we've written to storage. Also, 'Mutation' has now become 'Operation' to better reflect the fact that some operations do not mutate state when applied. Diffs - src/master/registrar.cpp ee16121035db21d966ee151483dd23cbc4a495c2 Diff: https://reviews.apache.org/r/18341/diff/ Testing --- make check Thanks, Ben Mahler
Re: Review Request 18370: Refactor Cluster::Master::start methods
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18370/#review35561 --- src/tests/cluster.hpp https://reviews.apache.org/r/18370/#comment66163 s/normal production/default/ ? src/tests/cluster.hpp https://reviews.apache.org/r/18370/#comment66165 Why this name change? I rather liked the former. - Vinod Kone On Feb. 25, 2014, 9:23 p.m., Charlie Carson wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18370/ --- (Updated Feb. 25, 2014, 9:23 p.m.) Review request for mesos, Benjamin Hindman, Ben Mahler, and Vinod Kone. Bugs: MESOS-1024 https://issues.apache.org/jira/browse/MESOS-1024 Repository: mesos-git Description --- Refactor Cluster::Master::start methods There are currently two overloads of the Cluster::Masters::start function. One takes a argument of AllocatorProcess* and the other does not. The AllocatorProcess* overload serves two purposes: 1) it allows an alternative implementation of AllocatorProcess to be passed (i.e. a mock) 2) it changes the destruction timing so that the passed in argument can outlive the master. Beyond that, the two functions are identical. This changes the parameter to be OptionAllocatorProcess* and allows all the logic to be in one method. The old function exists for back-compat but now simply forward by passing in None() to the other function. This is for two purposes: 1) reduce code duplication 2) position the code so that we can also optionally pass in a mock repaier. Review: https://reviews.apache.org/r/18370 Diffs - src/tests/cluster.hpp d1bf680ed3f6a0fb16af85a21409d653d44522ca Diff: https://reviews.apache.org/r/18370/diff/ Testing --- make check Thanks, Charlie Carson
Re: Review Request 18370: Refactor Cluster::Master::start methods
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18370/ --- (Updated Feb. 26, 2014, 7:56 p.m.) Review request for mesos and Vinod Kone. Changes --- assigning to myself (@vinodkone) for shepherding this. Bugs: MESOS-1024 https://issues.apache.org/jira/browse/MESOS-1024 Repository: mesos-git Description --- Refactor Cluster::Master::start methods There are currently two overloads of the Cluster::Masters::start function. One takes a argument of AllocatorProcess* and the other does not. The AllocatorProcess* overload serves two purposes: 1) it allows an alternative implementation of AllocatorProcess to be passed (i.e. a mock) 2) it changes the destruction timing so that the passed in argument can outlive the master. Beyond that, the two functions are identical. This changes the parameter to be OptionAllocatorProcess* and allows all the logic to be in one method. The old function exists for back-compat but now simply forward by passing in None() to the other function. This is for two purposes: 1) reduce code duplication 2) position the code so that we can also optionally pass in a mock repaier. Review: https://reviews.apache.org/r/18370 Diffs - src/tests/cluster.hpp d1bf680ed3f6a0fb16af85a21409d653d44522ca Diff: https://reviews.apache.org/r/18370/diff/ Testing --- make check Thanks, Charlie Carson
[jira] [Resolved] (MESOS-945) Show framework host name in the WebUI
[ https://issues.apache.org/jira/browse/MESOS-945?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Vinod Kone resolved MESOS-945. -- Resolution: Fixed commit f19db574d4e644beed88100b198354733706422d Author: Bernd Mathiske be...@mesosphere.io Date: Wed Feb 26 11:45:57 2014 -0800 Added hostname attribute to framework and displayed it in master's web ui. Review: https://reviews.apache.org/r/18361 Show framework host name in the WebUI - Key: MESOS-945 URL: https://issues.apache.org/jira/browse/MESOS-945 Project: Mesos Issue Type: Improvement Components: webui Reporter: Vinod Kone Assignee: Bernd Mathiske Labels: newbie Fix For: 0.19.0 It would be great if we could see the hostname of the framework that is currently registered with the master in the webui. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
Fwd: Path forward to C++11.
Forwarding from the right account to avoid Spam filters ... -- Forwarded message -- From: Dominic Hamon dha...@twitter.com Date: Wed, Feb 26, 2014 at 11:46 AM Subject: Re: Path forward to C++11. To: dev@mesos.apache.org What do you mean by 'C++11 support' in the compiler? g++4.7 supports some features, not all. I think we need to define the set of C++11 features we want to support first before making any decisions regarding timeline. Then we can look at the state of the nation in regards to C++11 feature support in different compilers on different OSs, and which are available out of the box. We don't want to limit our community involvement by shutting out certain platforms for development. Ideally, we'd be able to say something like 0.19.0 builds on g++4.7 and clang 3.3 or greater and 0.20.0 builds on g++4.8 and clang 3.4 or greater instead of blanket C++11 support. On Wed, Feb 26, 2014 at 10:20 AM, Benjamin Mahler benjamin.mah...@gmail.com wrote: Update on MESOS-750: 0.17.0 is being released with the --with-cxx11 capability at configure time. I chatted with some of the committers and we'd like to move forward with the final two phases of MESOS-750 as follows: 0.18.0: --with-cxx11 will become the _default_. --without-cxx11 will be available for those who wish to use older compilers. 0.19.0: c++11 required; Mesos will not build with a compiler that does not have c++11 support. If there are any objections please speak now. If not, I'll update the ticket to reflect the current path towards C++11. On Mon, Oct 28, 2013 at 8:09 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Thanks for the heads up, we're likely to upgrade to 3.3.6, but I've yet to confirm that it builds with c++11. On Thu, Oct 24, 2013 at 8:10 AM, Tim St Clair tstcl...@redhat.com wrote: +1 to the approach. FWIW made a quick update to the zookeeper.spec ( http://koji.fedoraproject.org/koji/buildinfo?buildID=465529) and updated: export CC='gcc -std=c11' export CXX='g++ -std=c++11' prior to 'configure', and it failed compilation pretty quickly. It may make sense to track integration issues and versions...somewhere. So when --with-cpp11 is enabled, it can hard check on version = fixed_version. Cheers, Tim - Original Message - From: Benjamin Mahler benjamin.mah...@gmail.com To: dev@mesos.apache.org Sent: Wednesday, October 23, 2013 4:46:49 PM Subject: Path forward to C++11. As discussed in MESOS-750 https://issues.apache.org/jira/browse/MESOS-750, we would like to move to C++11. The proposed plan is to do this in phases to allow everyone time to adjust to this change: Phase 1: Send an email to the dev@ list soliciting objections / feedback. We're currently here. :) Phase 2: Add an option --with-cpp11 that defaults to false. Phase 3: At this point, developers can and should begin using --with-cpp11. In this phase, we'll need to make sure the project and dependencies can compile with c++11. Phase 4: Default to --with-cpp11 but still leave the option in order to allow folks to turn this off if needed. At this point, we still cannot rely on C++11. Phase 5: Remove --with-cpp11 leaving no ability to turn it off. From this point forward, we can begin to use C++11 features unconditionally. Please reply to this thread with any objections, comments, or advice! To get a better sense of how others feel about this (I currently only know that some of the other core contributors are on board), please feel free to reply with something as simple as a +1 as well. Thanks! Ben -- Cheers, Tim -- Dominic Hamon | @mrdo | Twitter *There are no bad ideas; only good ideas that go horribly wrong.*
Re: Review Request 18341: Modified the Registrar to notify callers only _after_ the storage operation completes.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/#review35563 --- Ship it! src/master/registrar.cpp https://reviews.apache.org/r/18341/#comment66166 Does this have to be virtual? - Vinod Kone On Feb. 26, 2014, 7:54 p.m., Ben Mahler wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18341/ --- (Updated Feb. 26, 2014, 7:54 p.m.) Review request for mesos, Benjamin Hindman and Vinod Kone. Bugs: MESOS-764 https://issues.apache.org/jira/browse/MESOS-764 Repository: mesos-git Description --- We previously were setting the Promises in some cases _before_ writing to storage. While this approach was correct, it requires some time and explanation to convince oneself of this. This is why this patch simplifies the semantics by only setting the Promises _after_ we've written to storage. Also, 'Mutation' has now become 'Operation' to better reflect the fact that some operations do not mutate state when applied. Diffs - src/master/registrar.cpp ee16121035db21d966ee151483dd23cbc4a495c2 Diff: https://reviews.apache.org/r/18341/diff/ Testing --- make check Thanks, Ben Mahler
Build failed in Jenkins: Mesos-Trunk-Ubuntu-Build-Out-Of-Src-Set-JAVA_HOME #1919
See https://builds.apache.org/job/Mesos-Trunk-Ubuntu-Build-Out-Of-Src-Set-JAVA_HOME/1919/changes Changes: [niklas] Added new documentation page for frameworks. -- [...truncated 4703 lines...] I0226 19:18:08.511400 24627 master.cpp:2339] Queuing up authentication request from scheduler(1)@67.195.138.9:38323 because authentication is still in progress I0226 19:18:08.932227 24633 master.cpp:96] No whitelist given. Advertising offers for all slaves I0226 19:18:09.135534 24631 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 58907ns I0226 19:18:10.135767 24634 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 31688ns I0226 19:18:11.136059 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 50719ns I0226 19:18:12.137267 24633 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 32653ns I0226 19:18:13.137533 24629 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 30145ns W0226 19:18:13.511966 24629 sched.cpp:362] Authentication timed out I0226 19:18:13.512111 24629 sched.cpp:322] Failed to authenticate with master master@67.195.138.9:38323: Authentication discarded I0226 19:18:13.512151 24629 sched.cpp:267] Authenticating with master master@67.195.138.9:38323 I0226 19:18:13.512291 24627 authenticatee.hpp:128] Creating new client SASL connection I0226 19:18:13.512517 24634 master.cpp:2339] Queuing up authentication request from scheduler(1)@67.195.138.9:38323 because authentication is still in progress I0226 19:18:13.932324 24627 master.cpp:96] No whitelist given. Advertising offers for all slaves I0226 19:18:14.138712 24629 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 52245ns I0226 19:18:15.138974 24630 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 37203ns I0226 19:18:16.139312 24628 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 38662ns I0226 19:18:17.139726 24631 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 40628ns I0226 19:18:18.139964 24627 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 29378ns W0226 19:18:18.512387 24634 sched.cpp:362] Authentication timed out I0226 19:18:18.512657 24629 sched.cpp:322] Failed to authenticate with master master@67.195.138.9:38323: Authentication discarded I0226 19:18:18.512722 24629 sched.cpp:267] Authenticating with master master@67.195.138.9:38323 I0226 19:18:18.512908 24628 authenticatee.hpp:128] Creating new client SASL connection I0226 19:18:18.513178 24629 master.cpp:2339] Queuing up authentication request from scheduler(1)@67.195.138.9:38323 because authentication is still in progress I0226 19:18:18.932801 24632 master.cpp:96] No whitelist given. Advertising offers for all slaves I0226 19:18:19.141152 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 31871ns I0226 19:18:20.141477 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 29318ns I0226 19:18:21.141680 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 30149ns I0226 19:18:22.141899 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 29267ns I0226 19:18:23.142094 24632 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 28831ns W0226 19:18:23.513563 24632 sched.cpp:362] Authentication timed out I0226 19:18:23.513694 24632 sched.cpp:322] Failed to authenticate with master master@67.195.138.9:38323: Authentication discarded I0226 19:18:23.513725 24632 sched.cpp:267] Authenticating with master master@67.195.138.9:38323 I0226 19:18:23.513818 24632 authenticatee.hpp:128] Creating new client SASL connection I0226 19:18:23.513983 24632 master.cpp:2339] Queuing up authentication request from scheduler(1)@67.195.138.9:38323 because authentication is still in progress I0226 19:18:23.933203 24627 master.cpp:96] No whitelist given. Advertising offers for all slaves I0226 19:18:24.142500 24627 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 35792ns I0226 19:18:25.142745 24633 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 25794ns I0226 19:18:26.143082 24633 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 31622ns I0226 19:18:27.143302 24633 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 25548ns I0226 19:18:28.143506 24633 hierarchical_allocator_process.hpp:688] Performed allocation for 1 slaves in 26022ns W0226 19:18:28.513929 24633 sched.cpp:362] Authentication timed out I0226 19:18:28.514027 24633 sched.cpp:322] Failed to authenticate with master master@67.195.138.9:38323: Authentication discarded I0226 19:18:28.514053 24633 sched.cpp:267] Authenticating with master
Review Request 18532: Fix ROOT_CGROUPS_Write test.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18532/ --- Review request for mesos, Ben Mahler and Vinod Kone. Repository: mesos-git Description --- Fixed test by moving cgroups::create before fork so only the parent creates it. Also cleaned up the code paths for the parent/child. Diffs - src/tests/cgroups_tests.cpp 6a111e500b3824b87569d6014f966fb35e62ccd0 Diff: https://reviews.apache.org/r/18532/diff/ Testing --- make check # linux Thanks, Ian Downes
Jenkins build is back to normal : Mesos-Trunk-Ubuntu-Build-Out-Of-Src-Set-JAVA_HOME #1920
See https://builds.apache.org/job/Mesos-Trunk-Ubuntu-Build-Out-Of-Src-Set-JAVA_HOME/1920/changes
Re: Review Request 18532: Fix ROOT_CGROUPS_Write test.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18532/ --- (Updated Feb. 26, 2014, 9:55 p.m.) Review request for mesos and Vinod Kone. Changes --- assigned to myself (@vinodkone) for review. Repository: mesos-git Description --- Fixed test by moving cgroups::create before fork so only the parent creates it. Also cleaned up the code paths for the parent/child. Diffs - src/tests/cgroups_tests.cpp 6a111e500b3824b87569d6014f966fb35e62ccd0 Diff: https://reviews.apache.org/r/18532/diff/ Testing --- make check # linux Thanks, Ian Downes
Re: Review Request 18532: Fix ROOT_CGROUPS_Write test.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18532/#review35568 --- Ship it! Is there a bug that needs to be attached? src/tests/cgroups_tests.cpp https://reviews.apache.org/r/18532/#comment66175 align errno with write ? - Vinod Kone On Feb. 26, 2014, 8:19 p.m., Ian Downes wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18532/ --- (Updated Feb. 26, 2014, 8:19 p.m.) Review request for mesos, Ben Mahler and Vinod Kone. Repository: mesos-git Description --- Fixed test by moving cgroups::create before fork so only the parent creates it. Also cleaned up the code paths for the parent/child. Diffs - src/tests/cgroups_tests.cpp 6a111e500b3824b87569d6014f966fb35e62ccd0 Diff: https://reviews.apache.org/r/18532/diff/ Testing --- make check # linux Thanks, Ian Downes
Re: Path forward to C++11.
Being explicit about compiler version requirements sounds great! GCC is still our explicitly supported compiler toolchain and it's how we test releases for production purposes, at least at Twitter. We have been supporting Clang as a best effort so far, but I agree it would be great to document supported versions. Taking a look at some of our CI, I see Clang 3.3. is building Mesos. Looking at CI for gcc, I see the following building with C++11: *Mesos builds with C++11:* CentOS 5 - with devtoolset-2 gcc 4.8.1 CentOS 6 - with devtoolset-2 gcc 4.8.1 Fedora 19 - gcc 4.8.2 Fedora 20 - gcc 4.8.2 Ubuntu 13.10 - gcc 4.8.x *Mesos does not build with C++11:* CentOS 5 - gcc 4.1.2 CentOS 6 - gcc 4.4.7 Ubuntu 12.10 - gcc 4.7.x Ubuntu 12.04 - gcc 4.6.x So we already do not support gcc 4.7 for C++11. It's important here to make the distinction that this is for development, ideally those are running from RPMs are agnostic to the gcc version used. But, I suppose the c++11 shared library will not present on the machine. On Wed, Feb 26, 2014 at 12:08 PM, Dominic Hamon dha...@twopensource.comwrote: Forwarding from the right account to avoid Spam filters ... -- Forwarded message -- From: Dominic Hamon dha...@twitter.com Date: Wed, Feb 26, 2014 at 11:46 AM Subject: Re: Path forward to C++11. To: dev@mesos.apache.org What do you mean by 'C++11 support' in the compiler? g++4.7 supports some features, not all. I think we need to define the set of C++11 features we want to support first before making any decisions regarding timeline. Then we can look at the state of the nation in regards to C++11 feature support in different compilers on different OSs, and which are available out of the box. We don't want to limit our community involvement by shutting out certain platforms for development. Ideally, we'd be able to say something like 0.19.0 builds on g++4.7 and clang 3.3 or greater and 0.20.0 builds on g++4.8 and clang 3.4 or greater instead of blanket C++11 support. On Wed, Feb 26, 2014 at 10:20 AM, Benjamin Mahler benjamin.mah...@gmail.com wrote: Update on MESOS-750: 0.17.0 is being released with the --with-cxx11 capability at configure time. I chatted with some of the committers and we'd like to move forward with the final two phases of MESOS-750 as follows: 0.18.0: --with-cxx11 will become the _default_. --without-cxx11 will be available for those who wish to use older compilers. 0.19.0: c++11 required; Mesos will not build with a compiler that does not have c++11 support. If there are any objections please speak now. If not, I'll update the ticket to reflect the current path towards C++11. On Mon, Oct 28, 2013 at 8:09 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Thanks for the heads up, we're likely to upgrade to 3.3.6, but I've yet to confirm that it builds with c++11. On Thu, Oct 24, 2013 at 8:10 AM, Tim St Clair tstcl...@redhat.com wrote: +1 to the approach. FWIW made a quick update to the zookeeper.spec ( http://koji.fedoraproject.org/koji/buildinfo?buildID=465529) and updated: export CC='gcc -std=c11' export CXX='g++ -std=c++11' prior to 'configure', and it failed compilation pretty quickly. It may make sense to track integration issues and versions...somewhere. So when --with-cpp11 is enabled, it can hard check on version = fixed_version. Cheers, Tim - Original Message - From: Benjamin Mahler benjamin.mah...@gmail.com To: dev@mesos.apache.org Sent: Wednesday, October 23, 2013 4:46:49 PM Subject: Path forward to C++11. As discussed in MESOS-750 https://issues.apache.org/jira/browse/MESOS-750, we would like to move to C++11. The proposed plan is to do this in phases to allow everyone time to adjust to this change: Phase 1: Send an email to the dev@ list soliciting objections / feedback. We're currently here. :) Phase 2: Add an option --with-cpp11 that defaults to false. Phase 3: At this point, developers can and should begin using --with-cpp11. In this phase, we'll need to make sure the project and dependencies can compile with c++11. Phase 4: Default to --with-cpp11 but still leave the option in order to allow folks to turn this off if needed. At this point, we still cannot rely on C++11. Phase 5: Remove --with-cpp11 leaving no ability to turn it off. From this point forward, we can begin to use C++11 features unconditionally. Please reply to this thread with any objections, comments, or advice! To get a better sense of how others feel about this (I currently only know that some of the other core contributors are on board), please feel free to reply with something as simple as a +1 as well. Thanks! Ben -- Cheers, Tim --
Re: Path forward to C++11.
Thank you for the breakdown, that's really helpful. It looks like with 4.8.1 as the minimum supported version, we're looking very good for all features: http://gcc.gnu.org/gcc-4.8/cxx0x_status.html We may need to consider distributing the runtime library or statically linking it to allow users to run the binary from the rpm. On Wed, Feb 26, 2014 at 1:59 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Being explicit about compiler version requirements sounds great! GCC is still our explicitly supported compiler toolchain and it's how we test releases for production purposes, at least at Twitter. We have been supporting Clang as a best effort so far, but I agree it would be great to document supported versions. Taking a look at some of our CI, I see Clang 3.3. is building Mesos. Looking at CI for gcc, I see the following building with C++11: *Mesos builds with C++11:* CentOS 5 - with devtoolset-2 gcc 4.8.1 CentOS 6 - with devtoolset-2 gcc 4.8.1 Fedora 19 - gcc 4.8.2 Fedora 20 - gcc 4.8.2 Ubuntu 13.10 - gcc 4.8.x *Mesos does not build with C++11:* CentOS 5 - gcc 4.1.2 CentOS 6 - gcc 4.4.7 Ubuntu 12.10 - gcc 4.7.x Ubuntu 12.04 - gcc 4.6.x So we already do not support gcc 4.7 for C++11. It's important here to make the distinction that this is for development, ideally those are running from RPMs are agnostic to the gcc version used. But, I suppose the c++11 shared library will not present on the machine. On Wed, Feb 26, 2014 at 12:08 PM, Dominic Hamon dha...@twopensource.com wrote: Forwarding from the right account to avoid Spam filters ... -- Forwarded message -- From: Dominic Hamon dha...@twitter.com Date: Wed, Feb 26, 2014 at 11:46 AM Subject: Re: Path forward to C++11. To: dev@mesos.apache.org What do you mean by 'C++11 support' in the compiler? g++4.7 supports some features, not all. I think we need to define the set of C++11 features we want to support first before making any decisions regarding timeline. Then we can look at the state of the nation in regards to C++11 feature support in different compilers on different OSs, and which are available out of the box. We don't want to limit our community involvement by shutting out certain platforms for development. Ideally, we'd be able to say something like 0.19.0 builds on g++4.7 and clang 3.3 or greater and 0.20.0 builds on g++4.8 and clang 3.4 or greater instead of blanket C++11 support. On Wed, Feb 26, 2014 at 10:20 AM, Benjamin Mahler benjamin.mah...@gmail.com wrote: Update on MESOS-750: 0.17.0 is being released with the --with-cxx11 capability at configure time. I chatted with some of the committers and we'd like to move forward with the final two phases of MESOS-750 as follows: 0.18.0: --with-cxx11 will become the _default_. --without-cxx11 will be available for those who wish to use older compilers. 0.19.0: c++11 required; Mesos will not build with a compiler that does not have c++11 support. If there are any objections please speak now. If not, I'll update the ticket to reflect the current path towards C++11. On Mon, Oct 28, 2013 at 8:09 PM, Benjamin Mahler benjamin.mah...@gmail.comwrote: Thanks for the heads up, we're likely to upgrade to 3.3.6, but I've yet to confirm that it builds with c++11. On Thu, Oct 24, 2013 at 8:10 AM, Tim St Clair tstcl...@redhat.com wrote: +1 to the approach. FWIW made a quick update to the zookeeper.spec ( http://koji.fedoraproject.org/koji/buildinfo?buildID=465529) and updated: export CC='gcc -std=c11' export CXX='g++ -std=c++11' prior to 'configure', and it failed compilation pretty quickly. It may make sense to track integration issues and versions...somewhere. So when --with-cpp11 is enabled, it can hard check on version = fixed_version. Cheers, Tim - Original Message - From: Benjamin Mahler benjamin.mah...@gmail.com To: dev@mesos.apache.org Sent: Wednesday, October 23, 2013 4:46:49 PM Subject: Path forward to C++11. As discussed in MESOS-750 https://issues.apache.org/jira/browse/MESOS-750, we would like to move to C++11. The proposed plan is to do this in phases to allow everyone time to adjust to this change: Phase 1: Send an email to the dev@ list soliciting objections / feedback. We're currently here. :) Phase 2: Add an option --with-cpp11 that defaults to false. Phase 3: At this point, developers can and should begin using --with-cpp11. In this phase, we'll need to make sure the project and dependencies can compile with c++11. Phase 4: Default to --with-cpp11 but still leave the option in order to allow folks to turn this off if needed. At this point, we still cannot
Review Request 18540: assign bool directly to JSON::Value
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18540/ --- Review request for mesos and Benjamin Hindman. Repository: mesos-git Description --- assign bool directly to JSON::Value This change removes the workaround that was necessary now that https://issues.apache.org/jira/browse/MESOS-939 is committed. Review: https://reviews.apache.org/r/18540 Diffs - src/master/http.cpp e64093c2aa9d54ad615050ed0339ab0d6d4c6e41 src/tests/repair_tests.cpp ba6f50ad67ebe834ae4af5fc10bd39270ce91471 Diff: https://reviews.apache.org/r/18540/diff/ Testing --- make check Thanks, Charlie Carson
Re: Review Request 18540: assign bool directly to JSON::Value
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18540/#review35571 --- Ship it! Ship It! - Dominic Hamon On Feb. 26, 2014, 2:14 p.m., Charlie Carson wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18540/ --- (Updated Feb. 26, 2014, 2:14 p.m.) Review request for mesos and Benjamin Hindman. Repository: mesos-git Description --- assign bool directly to JSON::Value This change removes the workaround that was necessary now that https://issues.apache.org/jira/browse/MESOS-939 is committed. Review: https://reviews.apache.org/r/18540 Diffs - src/master/http.cpp e64093c2aa9d54ad615050ed0339ab0d6d4c6e41 src/tests/repair_tests.cpp ba6f50ad67ebe834ae4af5fc10bd39270ce91471 Diff: https://reviews.apache.org/r/18540/diff/ Testing --- make check Thanks, Charlie Carson
[jira] [Commented] (MESOS-890) Figure out a way to migrate a live Mesos cluster to a different ZooKeeper cluster
[ https://issues.apache.org/jira/browse/MESOS-890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13913694#comment-13913694 ] David Robinson commented on MESOS-890: -- I think option #1 is the better approach. Writing a heap of code to correctly handle option #2 seems like a lot of effort for something which should very rarely happen. IIRC the 75 second constraint can be adjusted in Aurora too. [~rgs], how difficult is #1 from the ZK perspective? I'm happy to test this in a dev/test cluster. Figure out a way to migrate a live Mesos cluster to a different ZooKeeper cluster - Key: MESOS-890 URL: https://issues.apache.org/jira/browse/MESOS-890 Project: Mesos Issue Type: Improvement Components: master Reporter: Raul Gutierrez Segales Assignee: Raul Gutierrez Segales I've been chatting with [~vinodkone] about approaching a live ZK cluster migration. Here are the options we came up with. For the descriptions we treat `zk1` as the current working cluster, `obs` as a bunch of ZooKeeper Observers [1] and `zk2` as the new cluster to which we need to migrate. Approach #1: Using Observers With this option we need to: * add obs to zk1 * restart slaves to have them use obs to find their master * restart the framework having it use obs to find the mesos master * restart the mesos masters having them use obs to perform their election * we then stop all ZK obs and remove their data (since they will need to sync up with an entirely new cluster, we need to lose the old data) * we restart ZK obs having them be part of zk2 * at this point the slaves, the framework and the masters can reach the ZK obs again and an election happens * optionally you can restart slaves, the framework and masters again using zk2 instead of the ZK obs if you wanted to decommission them. This assumes that we can do the last three steps in 75 secs (75 secs being the slave health check timeout). This is a reasonable assumption if the data size in zk2 is small enough to ensure that the ZK obs can sync up quickly with zk2. If zk2 is a new cluster with no data then this should be very fast. The good things of this approach are: * no mesos code change * it is very easy to rollback half way through, if need be The hard issues are: * Manipulating the ZK obs (i.e.: stopping, removing the data from zk1 and starting again) needs to be done with care. Messing up configs or not removing the data from zk1 on any of the ZK obs will cause problems * we need to restart all slaves to have them use the ZK obs instead of connecting to zk1 directly. But with slave recovery this isn't an issue, just an extra step. * same thing for the framework and the masters Approach #2: Dual publishing from mesos masters With this option we would augment the election handling code in mesos masters to have it deal with the notion of a primary and secondary ZK clusters. Master registration and election would then work as follows: * create an ephemeral|sequential znode in zk1 (i.e.: /path/to/znode/mesos_23) * create an ephemeral, but not sequential, znode in zk2 with the exact same path as what was created in zk1 (i.e.: /path/to/znode/mesos_23) * make sure both sessions, in zk1 and zk2, are always in the same state (i.e.: if one expires, the other one should be closed, etc.) For now, lets omit a few implementation details which might need extra care and assume we can make this work consistently in such a way that zk2 reflects accurately elections that happen in zk1. This means that regardless of being connected to zk1 or zk2, you always get the same master. Once we have this the migration steps would be: * restart slaves to have them use zk2 where masters can be found by virtue of what we implemented above * restart the framework so that it finds the mesos master in zk2 * stop all mesos masters (they all need to be stopped before moving to the next step) * start all mesos masters using zk2 as its primary and only cluster Again, this assumes we can do the last two steps in 75 secs (or if we needed to, we could bump the slave health check timeout). Which, again, sounds achievable given that masters have no state and their start-up time is very short. The good things of this approach are: - no tinkering with extra ZK servers nor with ZK configs The hard issues are: - extra code needs to be added to the election handling bits of mesos master to address a very rare, but probable, use-case of cluster migration. It might take a bit of time to get that code right. - it's easier to end up with a bad state if any of the mesos masters ends up with a bad config or is restarted earlier and ends up publishing differently than
Re: Review Request 18532: Fix ROOT_CGROUPS_Write test.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18532/ --- (Updated Feb. 27, 2014, 12:04 a.m.) Review request for mesos and Vinod Kone. Bugs: MESOS-1042 https://issues.apache.org/jira/browse/MESOS-1042 Repository: mesos-git Description --- Fixed test by moving cgroups::create before fork so only the parent creates it. Also cleaned up the code paths for the parent/child. Diffs - src/tests/cgroups_tests.cpp 6a111e500b3824b87569d6014f966fb35e62ccd0 Diff: https://reviews.apache.org/r/18532/diff/ Testing --- make check # linux Thanks, Ian Downes
Re: Review Request 18386: Option reference cleanup in mesos
On Feb. 24, 2014, 11:52 p.m., Ben Mahler wrote: src/linux/fs.hpp, line 63 https://reviews.apache.org/r/18386/diff/2/?file=500797#file500797line63 Well I see how you stumbled upon this fix.. :) Does this mean you did a grep to ensure all of the Option usage doesn't use the reference after the Option is re-assigned? Dominic Hamon wrote: I did not. I'll do that after posting this just to make sure (though all tests pass.. that means something, right? ;)) Let me know when you've done the search (might be good to spell this out in the 'Testing Done'). - Ben --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18386/#review35338 --- On Feb. 25, 2014, 12:10 a.m., Dominic Hamon wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18386/ --- (Updated Feb. 25, 2014, 12:10 a.m.) Review request for mesos and Ben Mahler. Bugs: MESOS-1008 https://issues.apache.org/jira/browse/MESOS-1008 Repository: mesos-git Description --- See summary Diffs - src/linux/fs.hpp 1d86dd0d24c3daae957b5eec387638d1e8e6d7db src/linux/fs.cpp e5f4f9a16becd4e5960d0cbb7f988736188b2426 src/log/log.cpp 62dc9286285d5981aa5fc63e125d3c3f51d4f457 src/sched/sched.cpp dcb3158d2acd93a107fda51b19e92899a092e628 Diff: https://reviews.apache.org/r/18386/diff/ Testing --- make check Thanks, Dominic Hamon
[jira] [Created] (MESOS-1045) Unrecognized file extension in CommandInfo.URI causes executor to exit
Vinod Kone created MESOS-1045: - Summary: Unrecognized file extension in CommandInfo.URI causes executor to exit Key: MESOS-1045 URL: https://issues.apache.org/jira/browse/MESOS-1045 Project: Mesos Issue Type: Bug Affects Versions: 0.18.0 Reporter: Vinod Kone Fix For: 0.18.0 Previously only files with known extensions (zip and [compressed] tar) were extracted. Since 0.18.0, unknown file extensions causes the fetcher to exit. This is a regression because 1) it is not the intended behavior per our documentation in mesos.proto and 2) it breaks Jenkins framework which uses a jar file with command executor. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
Review Request 18565: Fixed fetcher to ignore unrecognized file names.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18565/ --- Review request for mesos and Ian Downes. Bugs: MESOS-1045 https://issues.apache.org/jira/browse/MESOS-1045 Repository: mesos-git Description --- See bug for details. Diffs - src/launcher/fetcher.cpp 9c9f07d11b69fb08d435ea130f4bc53148ed7036 Diff: https://reviews.apache.org/r/18565/diff/ Testing --- make check Thanks, Vinod Kone
[jira] [Assigned] (MESOS-1045) Unrecognized file extension in CommandInfo.URI causes executor to exit
[ https://issues.apache.org/jira/browse/MESOS-1045?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Vinod Kone reassigned MESOS-1045: - Assignee: Vinod Kone Unrecognized file extension in CommandInfo.URI causes executor to exit -- Key: MESOS-1045 URL: https://issues.apache.org/jira/browse/MESOS-1045 Project: Mesos Issue Type: Bug Affects Versions: 0.18.0 Reporter: Vinod Kone Assignee: Vinod Kone Fix For: 0.18.0 Previously only files with known extensions (zip and [compressed] tar) were extracted. Since 0.18.0, unknown file extensions causes the fetcher to exit. This is a regression because 1) it is not the intended behavior per our documentation in mesos.proto and 2) it breaks Jenkins framework which uses a jar file with command executor. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (MESOS-1045) Unrecognized file extension in CommandInfo.URI causes executor to exit
[ https://issues.apache.org/jira/browse/MESOS-1045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13914106#comment-13914106 ] Vinod Kone commented on MESOS-1045: --- https://reviews.apache.org/r/18565/ Unrecognized file extension in CommandInfo.URI causes executor to exit -- Key: MESOS-1045 URL: https://issues.apache.org/jira/browse/MESOS-1045 Project: Mesos Issue Type: Bug Affects Versions: 0.18.0 Reporter: Vinod Kone Fix For: 0.18.0 Previously only files with known extensions (zip and [compressed] tar) were extracted. Since 0.18.0, unknown file extensions causes the fetcher to exit. This is a regression because 1) it is not the intended behavior per our documentation in mesos.proto and 2) it breaks Jenkins framework which uses a jar file with command executor. -- This message was sent by Atlassian JIRA (v6.1.5#6160)